Curiosity is one of the most important abilities we can keep alive.
It makes us ask questions. It pushes us to explore unfamiliar paths. It helps us notice problems and opportunities that other people may ignore.
But curiosity only becomes valuable when we turn it into action.
In the AI era, we now have tools that can help us investigate, compare, automate, and build much faster. An AI agent can analyse information, use software tools, write code, and interact directly with a browser.
This is incredibly powerful. But the goal should never be to let AI make us passive.
AI should make us more alert, more creative, and even more curious.
Curiosity Helps Us Find the Real Problem
When we receive a task, it is easy to start working immediately on the first solution that comes to mind.
As software engineers, we can be tempted to open our code editor too quickly. We think about the framework, the API, the database, or the architecture before making sure we clearly understand the real problem.
But the first interpretation is not always the correct one.
A request for a new dashboard may actually be a problem with unclear information. A request for a new feature may come from a confusing user journey. A repetitive process may continue for years simply because nobody has stopped to ask whether it can be automated.
This is where curiosity makes a big difference.
Before building, ask:
Why does this problem exist?
Who is experiencing it?
What is the expected outcome?
Is the current constraint real or only inherited from an old process?
Can we remove the problem instead of adding another layer of software?
Better questions lead to better understanding. Better understanding leads to better decisions. Better decisions lead to software that is genuinely useful.
Use AI to Create More Questions, Not Only Answers
Many people use AI like a traditional search engine. They ask one question, receive one response, and move on.
This is useful, but it is only a small part of what these systems can do.
Instead of asking an AI agent only to give you the answer, ask it to help you look at the problem from several directions.
For example:
Reframe this problem in five different ways. Show me the assumptions behind each version and give me the smallest experiment I can use to test them.
You can also ask:
What am I missing?
What would an experienced engineer challenge here?
What would a beginner notice that I now take for granted?
What is the simplest version of this solution?
What evidence would prove that my current idea is wrong?
The AI model will not always be right. We still need to analyse its responses and verify important information.
But it can rapidly increase the number of perspectives we consider.
This is how AI becomes a thinking partner rather than only an answer generator.
Connecting an AI Agent to a Browser
An AI model without access to tools can explain what you should do. An AI agent connected to a browser can help you actually investigate and complete the workflow.
The browser gives the agent access to the environment where a lot of our real work already happens.
It can open documentation, inspect several pages, follow links, compare information, fill repetitive fields, test a user journey, gather structured data, and check the visible result.
This is a major difference.
Imagine that you need to compare several software services. You could manually open every website, find the correct documentation, compare the features, and organise everything in a table.
Or you could give a browser-connected agent a clear objective and ask it to collect the same information for you.
The agent removes the repetitive part. You keep the important part: deciding which criteria matter and what the information means.
Browser automation can help us:
research unfamiliar topics across several reliable sources;
compare products, APIs, or technical approaches;
repeat browser-based quality-assurance checks;
collect and organise information from complex workflows;
identify steps that waste time;
prepare actions for human review;
verify whether a visible outcome actually happened.
This allows us to solve problems faster, but also gives us more time to think about the parts that genuinely require our judgement.
My Practical AI Curiosity Loop
I like to think about the process in five simple steps.
1. Ask
Start with the outcome you want, not the AI tool you want to use.
Do not begin with, โI need to automate this.โ
Begin with, โWhat am I trying to improve, and why does it matter?โ
Otherwise, you may spend time automating a process that should not exist in the first place.
2. Observe
Let the agent inspect the existing workflow before changing anything.
Ask it to map every step, identify repetition, find missing information, and show where human judgement is required.
For example:
Inspect this workflow in read-only mode. Explain every step and identify friction or repeated actions. Do not submit, publish, send, or delete anything.
This gives you a much clearer picture before you start building.
3. Explore
Ask for several possible approaches.
Do not automatically accept the first one. Compare the complexity, risks, maintenance, and expected value of each option.
Sometimes the best solution is a sophisticated automation. Sometimes it is a small script. Sometimes it is changing one confusing sentence in the interface.
4. Automate
Start with actions that are repetitive, reversible, and easy to verify.
Collecting information, organising data, preparing a draft, opening the correct pages, or running the same checks are good starting points.
The purpose of automation is not to create more activity. It is to make the outcome more reliable and give us back time for higher-value thinking.
5. Verify and Learn
An attempted action is not always a completed action.
If the agent fills a form, inspect the result. If it updates a page, reopen the page. If it gathers research, compare the most important claims with the original sources.
Then reflect on what you learned.
Did the agent reveal a hidden dependency? Was a step unnecessary? Did the original problem change once you could see the complete workflow?
This final step creates the curiosity for the next improvement.
Keep the Human in Control
Connecting an agent to a browser is powerful because it gives the AI the ability to act.
This is also why we need clear boundaries.
Reading a page is different from submitting a form. Preparing an email is different from sending it. Uploading a file is different from publishing it. Generating a change is different from proving that the change works.
The more important the action is, the more important human verification becomes.
I prefer to let an agent explore freely when the work is read-only and reversible. When an action affects another person, changes public information, spends money, or removes data, I want a clear confirmation and a visible final check.
Good automation does not remove responsibility. It makes responsibility clearer.
Thinking Differently Makes Us Better
Curiosity is not only a technical skill.
It can also make us better colleagues and better people.
When we stay curious, we become less attached to our first interpretation. We ask why somebody sees the problem differently. We try to understand the context and constraints around their decision.
This creates more humility and empathy.
AI can help here too. Instead of asking it to confirm what you already believe, ask it to challenge you constructively:
What is the strongest reasonable argument against my current position, and what evidence could change my mind?
This question can improve an architecture decision, a product discussion, a career choice, or even a personal disagreement.
Thinking differently does not mean changing your opinion every time. It means making sure your decision comes from a wider and clearer view.
A Simple Next Step
Choose one task you repeat regularly this week.
Do not try to automate everything immediately.
First, ask an AI agent to observe the task, map the workflow, and highlight what could be improved. Then choose one small, reversible step to automate.
It can take only 30 minutes. The important thing is to start, test, and review what you have learned.
Curiosity creates questions. Questions create experiments. Experiments create better ways of working.
AI can make us faster. But our curiosity makes sure we are moving in the right direction.
Use AI to automate what slows you down, explore what you do not yet understand, and keep yourself actively involved in every important decision.
Stay curious, keep building, and continue thinking differently ๐
Thanks for reading The Healthy Scientist! Subscribe for practical content about AI, software engineering, better workflows, and healthy ways of building.
I have built many open-source projects over the years. Visit my GitHub for inspiration or to contribute.
More engineering and AI content is also coming on my LinkedIn. Make sure to follow so you do not miss it! ๐ฅ










